Correlation analysis between fabric structure and color rendering of polyester colored spun woven fabric based on the improved relative discrimination criterion

To clarify the color variation and influencing factors between colored spun yarn and its fabrics with different weave structures, a multi-dimensional spectral feature extraction model integrating distance and shape features is established in this paper. Considering the spectral reflectance of colored spun yarn and its fabric fluctuates randomly, a relative discrimination criterion based on the category separable ratio model is proposed; in addition, the principal component analysis algorithm is applied to the correlation analysis of the colored spun product’s color change. The results show that the established multi-dimensional spectral feature extraction model and relative discrimination criteria can effectively characterize the spectral features of colored spun yarns and fabrics. The difference of spectral features between the plain fabric samples and colored spun yarn samples is the largest, the twill structure is the second, and the difference of satin structure is the smallest. For colored spun fabric samples, the spectral feature fluctuation of the satin structure is the highest; the plain weave is the second, while the twill is the lowest. In addition, the increase in twist coefficient leads to a decrease in the difference of spectral features, but the color variation remains the same. Besides, there is no significant difference in spectral features between yarn and fabric samples using the principal component analysis algorithm. In the visible light range, the high-dimensional spectral features are more suitable for the characterization and analysis of the colored textiles’ color information.